• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Continue searching

The effectiveness of art therapy in reducing symptoms of trauma, anxiety, and stress: A meta-analysis

Author: Emma R. Campbell
Art therapy is an established therapeutic modality used by many mental health practitioners today, and yet despite its growing popularity, concern has often been raised in response to the lack of empirically supported research regarding its effectiveness. In this study a meta-analysis was performed on both published and unpublished art-based intervention studies in order to find both an overall effect size (ES) and moderating factors that impact the outcome of art therapy on anxiety-related symptoms in clients. This meta-analysis included 24 studies and found art therapy to have a moderate overall ES of 0.53 (with a 95% confidence interval (CI) of 0.36 to 0.71) for reducing anxiety symptoms. As this analysis included treatment studies regardless of their overall stated intervention goal (as long as an objective measure of anxiety was reported), when this author examined only those studies that specifically aimed to reduce anxiety (versus improve psychological wellbeing, elevate mood along with reduce anxiety, or reduce posttraumatic stress disorder (PTSD) symptom severity) the ES was higher at 0.81 (n = 6; with a 95% CI of 0.52 to 1.10), which is within the range of other recent meta-analytic reviews of psychotherapeutic interventions for anxiety (Hirai & Clum, 2006; Manzoni, Pagnini, Castelnuovo, & Molinari, 2008; Mitte, 2005; Reger & Gahm, 2009). The author discusses clinically relevant issues related to the use of art therapy for anxiety and presents suggestions for further research.

Table of Contents

Abstract v Introduction 1 Research Hypotheses 6 Method 7 Selection of Studies 7 Studies Included and Study Characteristics 9 Measure of Treatment Effect 11 Results 13 Overall Effect Size 13 Analysis of Heterogeneity and Moderating Variables 16 Research Hypotheses 17 Discussion 20 Overall Effectiveness of Art Therapy in Reducing Anxiety Symptoms 20 Random Variation 21 Moderating Variables 21 Suggestions for Future Research 25 References 26 Appendices A: Journals Included in the Manual Literature Search 31 B: Descriptive Summary of Meta Regression Results 32 C: Coding Sheet 33

Art Therapy Meta-Analysis 1

The Effectiveness of Art Therapy in Reducing the Symptoms of Trauma, Anxiety, and Stress: A Meta-Analysis The need for effective, short-term, and cost-effective treatments for anxiety-related disorders is evidenced in the prevalence rates estimated by the Surgeon General’s latest report (U.S. Dept of Health and Public Services, 1999). In their epidemiological studies, the U.S. Department of Health and Public Services found that almost one in every six adults experiences a cluster of anxiety-related disturbances, to the point that their ability to function has been significantly impaired. The financial burden of this condition is significant, for it is estimated that when considering mental health services received per year (for all mental illnesses), direct costs added up to $69.0 billion in the year 1996 alone (U.S. Dept of Health and Public Services, 1999). As psychological research continues, the ever-pressing trend is to develop more cost-effective interventions with better treatment outcomes and more culturally sensitive strategies. Although there are treatment interventions that have proven to be effective with anxiety-related symptoms, it is appropriate for responsible researchers and clinicians to continually work on advancing both established and developing treatment interventions. This study, therefore, follows in a long line of studies attempting to map out new and uncharted territory. The modality of art therapy is a growing and intriguing field in mental health that requires further investigation into its potential assets for clientele. Artistic expression, through various means, has long been a human passion and diversion. It can hardly be surprising, therefore, when those in the therapeutic profession began to see its potential and usefulness in the healing process. Art therapy, as it is known today, is an established

Art Therapy Meta-Analysis 2

therapeutic modality in mental health and has, since its birth in the 1940s, been devoted to the idea that the process of creating art is “healing and life enhancing and is a form of nonverbal communication of thoughts and feelings” (Malchiodi, 2003, p.1). As the field of art therapy is still relatively new, there remains debate about what it actually is. There are those who propose that art is an adjunctive tool in the therapeutic process, and that it can be used as a means of enhancing clients’ ability to express their thoughts, feelings, and beliefs to the therapist. Others would claim that art is therapeutic in and of itself, and that it is the creative process of making art that is life enhancing. It is proposed, however, that the effectiveness of art therapy is derived from both of these perspectives and approaches (Malchiodi, 2003). But as with other effective psychological interventions, why is it that art therapists, along with many school counselors, psychologists, psychiatrists, and nurses, have turned toward this alternative approach? Those in the profession of clinical psychology often have found it helpful to assist their clients in externalizing or objectifying their personal struggles. This externalization can be particularly important in anxiety disorders in which the ability to objectively encounter imagery of phobias or trauma may aid the healing process (Chapman, Morabito, Ladakakos, Schreier, & Knudson, 2001). Advocates of art therapy suggest that it can achieve this goal (Chapman et al., 2001; Carlson, 1997). By creating a work of art that symbolizes clients’ thoughts, feelings, and struggles, clients are able to step back and evaluate their lives. The symbolism is suggested to create a safety net, for as the clients explain their art to the clinician the clients can speak as objective observers and begin to take ownership not only over their actual art work, but also of their issues represented therein.

Art Therapy Meta-Analysis 3

The ability to aid in externalizing issues is not the only attribute of this modality, for several other advantages have been noted throughout the years. For instance, it has been argued that art therapy is useful in its ability to reach across cultural boundaries (Wadeson, 2000), since embedded in each culture there lies some form of art. As art therapy is (for these purposes) a visual medium, no verbal dialogue is required for its benefits to be seen. An example of this can be seen in Bien’s (2005) research which found that art therapy added to the effectiveness of treatment given to anxious and depressed urban HIV-positive Native Americans who had become distrustful of the psychotherapeutic process. In addition to this enviable benefit, social support has been observed to increase in art therapy groups while anxiety is lessened due to the distraction the process can bring (Mapp & Koch, 2004; Nainis et al., 2006). Lastly, art therapy could also be seen as a cost-effective intervention as it can readily be used in larger formats, such as classrooms, groups, and community events (Banks, 1996; Liebmann, 1986; Mapp & Koch, 2004). The American Art Therapy Association (AATA) was formed in 1969 with Dr. Myra Levick as its first president. Since its conception, the association has sought to provide clarity to the field, support its members, and create standards of professional competence. Currently, there are 33 AATA approved graduate schools providing masters degrees in art therapy. With worldwide interest, the United States is far from being unique in its development of this therapeutic intervention. The International Networking Group of Art Therapists, for instance, is an organization developed to facilitate the communication of art therapists and others interested in art therapy from almost 80 different countries

Art Therapy Meta-Analysis 4

around the world. There are also organizations similar to the AATA in other countries, such as Britain, Australia, Germany, Ireland, and Canada (AATA, 2006). Art therapists have found the use of art to be helpful with individuals of all ages as well as within groups, families, and couples (Armstrong, 2002; Gladding & Newsome, 2003; Hartz & Thick, 2005; Mapp & Koch, 2004; Riley, 2003; Wald, 2003), and yet concern still remains that art therapy lacks an adequate research base. Many, even within the field, have expressed their desire to see more empirically validated and experimentally designed studies (Art Therapy, 2003; Kaplan, 2005). Some say that research is not a natural inclination for advocates of art therapy, while others say that the field does not lend itself to traditional models of research (Deaver, 2002; Kaplan, 2000). Regardless of the reasoning, many supporters feel that, although they believe in the effectiveness of the modality, they (as a field) are doing an inadequate job of promoting its effectiveness (Bellmer, Hoshino, Schrader, Strong, & Hutzler, 2003). Further, there lies the question of efficacy versus effectiveness in art therapy treatment outcomes. An analysis of efficacy, as delineated by Aveline, Strauss, and Stiles (2007), would entail the exclusive examination of randomized trials which controlled for specific clinical populations and presented a highly manualized treatment approach. This research model would provide a high standard of internal validity, but would often fail at achieving significant external validity, or generalizability. An examination of effectiveness, however, assesses the applicability of a particular therapeutic approach in routine clinical populations, in which participants are not as highly controlled for diagnoses (meaning, comorbid conditions can occur) and the treatment approach is not typically manualized. The effectiveness model presents with the opposite attributes of an

Art Therapy Meta-Analysis 5

efficacy model, in that the results are usually high in generalizability and low in internal validity (Aveline, Strauss, & Stiles, 2007). Although a lower level of internal validity is a limitation with any assessment of treatment outcomes, the nature of art therapy interventions (i.e., its complexity and variability across settings and clinicians) along with the available breadth of art therapy research, demonstrates that an examination of effectiveness, rather then efficacy, is warranted. Therefore, despite modest gains in research and empirical support, art therapy remains an infant surrounded by the older adults of psychological theory and intervention. Research continues to be generated; however, more often than not arguments for the effectiveness of art therapy rely on, at best, a few experimental studies, and, at worst, anecdotal evidences. In the only other systematic review of art therapy known to this author, for instance, the authors (Reynolds, Nabors, & Quinlan, 2000) identified the available “published empirical evidence regarding art therapy effectiveness” and offered “outcome trends” based on a qualitative review of the literature. As significant and valuable as this type of qualitative review can be, the study was unable to produce a standardized metric that could objectively compare the effectiveness of art therapy to other therapeutic endeavors. Moreover, after conducting a survey of 1,000 psychology professors active in the American Psychological Association (APA) it was found that art therapy was seen in a negative light, thus challenging those in support of its practice to both better communicate its effectiveness and to supply further and more powerful empirical treatment outcomes (Bellmer et al., 2003). In light of this challenge, the purpose of this study was to conduct a meta-analysis on the therapeutic

Art Therapy Meta-Analysis 6

effectiveness of art therapy, providing a better awareness of the field’s contributions and limitations in relation to anxiety-related symptomatology. Research Hypotheses For this present study, the following research hypotheses guided the analysis: 1. Is art therapy an effective treatment for anxiety-related symptoms? It was hypothesized that the overall average effect size for art therapy would be both positive and significantly greater than zero, demonstrating the moderate effectiveness of the intervention. This assumption was based on the recent findings in the field which have appeared to demonstrate the effectiveness of this treatment (e.g., Puig, Lee, Goodwin, & Sherrard, 2006; Mallay, 2002). 2. Do children benefit more from art therapy than those in other age groups? In Driessnack’s (2005) meta-analytic review of children’s drawings during interviews, Driessnack found that the simple act of allowing a child to draw while being interviewed facilitated increased communication. Driessnack (2005) explained that the addition of art aided in the elicitation of the children’s voices, “whose command of the language and comfort in the health-care system is limited” (p. 415). Based on Driessnack’s (2005) proposal that art provides a significant communication tool for children, it was predicted that the effect size for children (up to age 17) with this type of intervention would be significantly greater than for adults (ages 18+). 3. Is art therapy more effective with those who suffer from posttraumatic stress symptoms (as opposed to other anxiety-related conditions)? It was predicted that there would be a greater effect size for participants presenting with posttraumatic

Art Therapy Meta-Analysis 7

stress symptoms than for those who presented with other anxiety-related symptoms due to the recent research that has shown art therapy’s effectiveness in improving the symptoms of PTSD with survivors of war (e.g., Baker, 2006; Spinner, 2007; van der Velden, 2005). 4. Will those participants who suffered a mass trauma (e.g., combat, natural disaster, etc.) benefit more from a group format for art therapy than an individual format? It was hypothesized that, for those participants whose PTSD symptoms were related to a mass trauma, the effect size for the group format would be greater than the effect size for the individual format. This assumption was based on the idea that the participants would have not only the benefit of the trained administrator in listening and dialoguing about their art, but the other group participants as well, who have gone through similar experiences (related to the mass trauma). 5. Does the length of treatment influence the effectiveness of the art therapy intervention? It was hypothesized that the longer the duration of treatment, the greater the benefit of art therapy for the participant. Therefore, it was anticipated that the treatment duration will be positively correlated with the magnitude of the effect sizes, which is based on the idea that the longer a treatment the greater the effectiveness of the intervention (e.g., Seligman, 1995). Method Selection of Studies Publication bias is a common criticism of meta-analytic reviews that choose to survey only published studies (Glass, McGaw, & Smith, 1981). In short, studies which

Art Therapy Meta-Analysis 8

are published tend to have more significant findings, larger participant sizes, and higher power than those that are passed up for publication. Reviews of such limited literature would likely skew the appropriate results in favor of the treatment modality. For that reason I have chosen to gather both published and unpublished studies (dissertations). Both types of studies were found through various search strategies. Electronic databases (PsycINFO, PsycARTICLES, ERIC, and MEDLINE) were searched through the date of April 3, 2009, and the search parameters were established based on the following keywords: art therapy, therapeutic art, art in therapy, creative therapy, and expressive therapy, all in combination with the words anxiety or stress. Additionally, three journals that regularly print articles on art therapy (see Appendix A for dates and list of journals) were manually browsed, and on April 1, 2009, this author searched through AATA’s website, which posted additional outcome studies of art therapy. Furthermore, studies found through the above methods were scanned for other relevant studies to include in the present meta-analysis. In order to be included in the meta-analytic review articles had to be written in English and meet the following criteria: (a) employment of either a control group or a single group pretest/posttest research design, (b) adequate data for calculating effect size and inverse variance weight (see explanation below), (c) use of art therapy as the primary intervention (studies which included additional therapeutic interventions, viewed as confounding variables, were excluded), and (d) administration of a valid, reliable, and objective measure of anxiety. The inverse variance weight for a meta-analysis refers to the inverse of the squared standard error value, as a Hedges’ g uses the standard error rather than the sample size to compute the weights for individual effect sizes (Lipsey &

Art Therapy Meta-Analysis 9

Wilson, 2001). Studies were excluded when they used a case study methodology, and only journal articles and unpublished dissertations were considered. For this study, an art therapy intervention was assigned when either the participants were simply asked to create a tangible form of visual art (painting, drawing, pottery, etc.) or the intervention integrated a form of visual art with a model of counseling or psychotherapy. This meta- analysis did not include other versions of arts in psychotherapy, such as dance/movement, drama, music, or poetry therapies. Studies Included and Study Characteristics After a systematic search protocol, 19 studies were found to meet inclusion criteria for this meta-analysis. Included in this total were 17 journal articles and two doctoral dissertations. Unfortunately, as articles and dissertations were gathered to assess their inclusion viability, three studies (one journal article and two dissertations; Asawa, 2004; Lawry, 1997; Pergamenter, 1982) were unable to be retrieved by this institution’s inter- library loan department, therefore it is unclear if these would have ended up in the final list of studies. Of the studies included, however, five had two separate art therapy treatment groups (Curry & Kasser, 2005; Henderson, Rosen, & Mascaro, 2007; Lyshak- Stelzer, Singer, St. John, & Chemtob, 2007; Miller, 1993; Webb-Ferebee, 2003), which were each subsequently analyzed as separate studies. This brought the grand total of studies for this project up to 24. Each study was coded for moderating variables by two raters (this author and an advanced doctoral student), with an initial agreement of 96%. Discrepancies were discussed among the raters, re-evaluated, and a final decision was made by this author. The various study characteristics coded are listed below (see Appendix C for coding sheet):

Art Therapy Meta-Analysis 10

1. Overall Stated Goal for Intervention (reduce anxiety; elevate mood or elevate mood and reduce anxiety; improve general psychological functioning/wellbeing, or reduce PTSD symptom severity) 2. Type of Outcome Measure Used (anxiety, PTSD symptom severity, or both) 3. Name, and number of anxiety or PTSD outcome measures (Type: Beck Anxiety Inventory (BAI); Behavior Assessment System for Children, Self-Report of Personality (BASC-SRP); Hospital Anxiety and Depression Scale, Anxiety Score (HADS); Posttraumatic Stress Diagnostic Scale (PDS); Profile of Mood States, Tension-Anxiety Score (POMS); Revised Children’s Manifest Anxiety Scale (RCMAS); State Anxiety Inventory (SAI); Symptom Checklist-90-Revised, Anxiety Scale (SCL-90-R); State Trait Anxiety Inventory (STAI); Trauma Symptom Child Checklist (TSCC); or UCLA PTSD Reaction Index) (Number: continuous variable) 4. Reason for Anxiety (induced for study; medical diagnosis; psychological diagnosis (not PTSD); PTSD; related to another person’s medical diagnosis, disability or death; or other) 5. Theoretical Stance of Investigator (art as therapy or art psychotherapy) 6. Publication Status (published journal article or unpublished dissertations) 7. Study Design (experimental, quasi-experimental, or single group pre-test/post- test) 8. Type of Control Used (placebo, treatment as usual, no treatment, wait-list, or N/A: single group pre-test/post-test design) 9. Treatment Setting (hospital, university, other/misc., or unknown)

Art Therapy Meta-Analysis 11

10. Treatment Format (group, individual, or unclear) 11. Treatment Duration (continuous variable coded in weeks of treatment) 12. Demographics of Participants a. Sample Size (continuous variable) b. Gender (continuous variable: percentage of females) c. Age Group (children: ages 7-17, adults: ages 18+, or mixed) d. Mean Age (continuous variable) e. Ethnicity (continuous variable: percentage of participants who were European American, African American, Latino American, Asian American, Native American, Biracial, International, or Minority: non- white) Measure of Treatment Effect In order to assess the effectiveness of art therapy on anxiety symptoms, a meta- analysis was performed. Meta-analyses, which have grown in popularity since the late 1970s, provide a researcher with the ability to summarize large amounts of quantitative data from a diverse array of studies (Lipsey & Wilson, 2001). With a meta-analysis, a researcher transforms the outcome results of a study into a standardized metric known as an effect size, which is the difference in posttest means for the experimental and control groups divided by the pooled standard deviation. Once an effect size is determined for each qualifying study, an overall weighted effect size for the entire group is computed. Furthermore, this style of investigation provides for the coding and analysis of moderating factors, to assess their potential impact on the computed effect size.

Art Therapy Meta-Analysis 12

The effect size value for this meta-analysis was Hedges’ g (Hedges and Olkin, 1985). As stated above, the approach in a meta-analysis is to weight individual effect sizes from each study. Some meta-analyses will perform this weight based solely on sample size, but when using the Hedges’ g, the weights are computed by considering both the sample size and by using the standard error of the effect size itself (Lipsey & Wilson, 2001). Therefore, in conducting this meta-analysis, each qualifying study had to provide not only a sample size, but also an adequate amount of data to produce both an effect size value and the inverse of the squared standard error, or what is known as the inverse variance weight (Lipsey & Wilson, 2001). The software used to conduct this meta- analysis, Comprehensive Meta Analysis (Borenstein, Hedges, Higgins, & Rothstein, 2008, version 2.2.048), states that in order to yield a Hedges’ g, first a standardized mean difference (d) is computed. The formula for d is: d = m e – m c

s p

where m e is the posttest mean of the experimental group, m c is the posttest mean of the comparison group, and s p is the pooled standard deviation of both groups. The program then multiples the product, d, by the factor J: J = 1 – 3

4*df – 1 (where df is the total sample size of each group minus the number of groups) to correct for bias and yield a Hedges’ g (Borenstein et al., 2008). When studies did not present posttest data in the format listed above (see d), other metrics of measurement (such as, t) were used to calculate the same Hedges’ g. In some cases, a single study produced more than one effect size, due to its use of two or more anxiety and/or PTSD outcome

Art Therapy Meta-Analysis 13

measures. When this occurred in the present study, then the multiple effect sizes for the particular study were automatically averaged to produce a single effect size, which was then the datum used for the rest of the analysis. Overall, this study was conducted in compliance with the outlined procedures given by Lipsey and Wilson (2001) and Hedges and Olkin (1985). Results Overall Effect Size The overall effect size for the 24 studies analyzed was 0.53 (with a 95% CI of 0.36 to 0.71), and was highly significant (p < 0.001). Using standard interpreting procedures (Durlak, 1995), an effect size around 0.20 is considered to be small, while one around 0.50 is seen as moderate, and an effect size of around 0.80 or higher is labeled as large. Thus, overall it can be concluded that art therapy yields a moderate effect size. This result, however, summarizes a set of studies that included a range of different overarching goals for the art therapy intervention. When this author examined only those studies that stated an overall intervention goal of reducing anxiety (as opposed to: a) reducing PTSD symptom severity; b) elevating mood, or elevating mood and reducing anxiety; or c) improving psychological functioning/wellbeing), the resulting effect size was 0.81 (n = 6, with a 95% CI of 0.52 to 1.10, p < 0.001), demonstrating a large effect size. It is proposed by this author that the true effect size of art therapy on reducing symptoms of anxiety is more similar to this last result of 0.81, than the former result (of 0.53). When calculating the overall effect size (0.53), the majority of the studies examined (75%, n = 18) did not state that their primary goal for the art therapy intervention was to reduce anxiety. Rather, the majority of the studies included in this

Art Therapy Meta-Analysis 14

meta-analysis (75%, n = 18) sought to improve the participants’ general psychological functioning/wellbeing, reduce their PTSD symptom severity, or elevate their mood along with reducing their anxiety. These latter goals appear to be peripheral to the task of simply reducing anxiety alone and seem to have substantially influenced the effectiveness of the art therapy intervention in accomplishing the goal of reducing anxiety. Therefore, as mentioned above, it is proposed by this author that the true effectiveness of art therapy in reducing anxiety symptoms alone is more similar to 0.81, as it was gained exclusively from studies that sought to primarily accomplish that goal (of reducing anxiety symptoms). For a descriptive summary of the effect size results see Table 1 and Appendix B. Table 1 Descriptive Summary of Effect Size Results No. of Effect p-value Defining Parameter studies size 95% CI (2-tail)

All Studies 24 0.53 0.36 – 0.71 < 0.001

Overall Stated Goal for Intervention a

Anxiety (Reduce Anxiety) 6 0.81 0.52 – 1.10 < 0.001 PTSD (Reduce Symptoms) 6 0.52 0.19 – 0.85 0.002 Mood (Elevate Mood & Reduce Anx) b 5 0.61 0.14 – 1.08 0.010 Psych. Wellbeing (Improve) b 7 0.23 0.03 – 0.43 0.023

Reason for Anxiety c

Induced for Study 3 1.30 0.96 – 1.64 < 0.001 Medical Diagnosis 3 -0.01 -0.38 – 0.37 0.983* Psychological Diagnosis (not PTSD) 3 0.46 0.14 – 0.79 0.005 PTSD 8 0.42 0.19 – 0.66 < 0.001 Another’s Medical Dx or Death 5 0.57 0.27 – 0.86 < 0.001 Other 2 0.74 0.26 – 1.21 0.002

PTSD versus Other Anxiety-Related Conditions Anxiety 16 0.59 0.35 – 0.82 < 0.001 PTSD 8 0.42 0.19 – 0.66 < 0.001

Art Therapy Meta-Analysis 15

Theoretical Stance Art as Therapy 13 0.63 0.39 – 0.88 < 0.001 Art Psychotherapy 11 0.40 0.17 – 0.63 0.001

Design Experimental 8 0.64 0.20 – 1.09 0.005 Quasi-Experimental 2 0.74 0.26 – 1.21 0.002 Single Group Pre/Posttest 14 0.47 0.27 – 0.66 < 0.001

Type of Control Used d

No Treatment 2 0.74 0.26 – 1.21 0.002 Placebo 4 0.92 0.05 – 1.78 0.038 Treatment as Usual 3 0.31 -0.01 – 0.63 0.055* Wait-list 1 0.50 -0.13 – 1.12 0.119*

Treatment Setting Hospital 12 0.45 0.18 – 0.73 0.001 University 6 0.74 0.32 – 1.16 < 0.001 Other/Misc. 4 0.51 0.27 – 0.75 < 0.001 Unknown 2 0.59 0.17 – 1.02 0.006

Treatment Format Group 17 0.59 0.37 – 0.80 < 0.001 Individual 6 0.36 0.02 – 0.71 0.040 Unclear 1 0.66 0.02 – 1.31 0.045

Age Group Adults (ages 18+) 16 0.55 0.32 – 0.77 < 0.001 Children (up to age 17) 7 0.49 0.20 – 0.78 0.001 Mixed 1 0.66 0.02 – 1.31 0.045

Study Type Published Journal Article 21 0.55 0.36 – 0.75 < 0.001 Unpublished Dissertation 3 0.35 0.01 – 0.70 0.042

Note. CI = confidence interval. Anx = anxiety. Dx = diagnosis. a The effectiveness of art therapy between characteristics in this parameter was significantly different. (Q B = 11.19, p = 0.011, using the mixed effects model.) b The effect sizes shown are still those that demonstrate art therapy’s effect on reducing anxiety-related symptoms, and not its effectiveness in either elevating mood or improving psychological wellbeing. c The effectiveness of art therapy between characteristics in this parameter was significantly different. (Q B = 24.60, p < 0.001, using a fixed effects model.) Medical Diagnosis was removed due to insignificance, though, when added to the analysis, this parameter remained significantly different. d Studies with a single group pretest/posttest design were not included under this parameter. *Italicized ES values indicate insignificant (p>.05) results.

Art Therapy Meta-Analysis 16

Analysis of Heterogeneity and Moderating Variables When conducting a meta-analysis, it is important to assess whether the individual effect sizes, used to produce the overall weighted effect size, represent the same population. In a homogeneous distribution, effect sizes differ only by what would be expected from sampling error, while a heterogeneous distribution demonstrates that there are further differences than can be explained through sampling error. The statistic used to determine heterogeneity is Q. According to Lipsey and Wilson (2001, p. 115), the Q statistic is “distributed as a chi-square with k – 1 degrees of freedom where k is the number of effect sizes (Hedges & Olkin, 1985).” Therefore, if Q is significant and exceeds the critical value for a chi-square of k – 1 degrees of freedom, then the null hypothesis of homogeneity must be rejected. When this occurs and the overall effect size distribution is determined to be heterogeneous, then further analysis is generally warranted (Lipsey & Wilson, 2001). In this present study, the overall effect size yielded a significantly heterogeneous distribution (Q = 74.40, p < 0.001). In assessing the excess variation detected by the Q statistic, an additional component to the analysis and computation arises. With a heterogeneous distribution, a researcher must determine if their sample of studies necessitates a fixed effects model, a random effects model, or a mixed effects model. Fortunately, the Q statistic of heterogeneity examines the utility of a fixed effects model (Lipsey & Wilson, 2001). A fixed effects model is used when the random variations between and within subgroups are assumed to derive only from sampling error. But when a distribution is determined to be heterogeneous the researcher is faced with one of three options: (1) continue to use the fixed effects model, but assume that the variability can be accounted for by the study

Full document contains 40 pages
Abstract: Art therapy is an established therapeutic modality used by many mental health practitioners today, and yet despite its growing popularity, concern has often been raised in response to the lack of empirically supported research regarding its effectiveness. In this study a meta-analysis was performed on both published and unpublished art-based intervention studies in order to find both an overall effect size (ES) and moderating factors that impact the outcome of art therapy on anxiety-related symptoms in clients. This meta-analysis included 24 studies and found art therapy to have a moderate overall ES of 0.53 (with a 95% confidence interval (CI) of 0.36 to 0.71) for reducing anxiety symptoms. As this analysis included treatment studies regardless of their overall stated intervention goal (as long as an objective measure of anxiety was reported), when this author examined only those studies that specifically aimed to reduce anxiety (versus improve psychological wellbeing, elevate mood along with reduce anxiety, or reduce posttraumatic stress disorder (PTSD) symptom severity) the ES was higher at 0.81 (n = 6; with a 95% CI of 0.52 to 1.10), which is within the range of other recent meta-analytic reviews of psychotherapeutic interventions for anxiety (Hirai & Clum, 2006; Manzoni, Pagnini, Castelnuovo, & Molinari, 2008; Mitte, 2005; Reger & Gahm, 2009). The author discusses clinically relevant issues related to the use of art therapy for anxiety and presents suggestions for further research.